On Tuesday, 7 August 2012 at 06:31:46 UTC, Brad Anderson wrote:
On Mon, Aug 6, 2012 at 11:59 PM, José Armando García Sancio <
[email protected]> wrote:
[snip]
Personally I am not a huge fan of using strings as a way to pass a function into a high-order function. I suspect that this string stuff became popular because D didn't have lambda declarations and type inference when the module was designed and implemented. I wonder if the module signatures would look different if you designed it to use
the current features of D.

To be honest, when first saw 'some_array.map!"a+1"()' a few years back when I first learned about D my knee jerk reaction was: "from where in
the world is 'a' coming from?". I remember I had to read
std/algorithm.d and whatever module implements unary "string"
functions to figure out what was going on. Anyways, I digress.

Part of me looks forward to the day that I can use D reliably another part of my wants to just wait for Rust to hopefully get it "right".

Thanks,
-Jose


There was a pull to switch std.algorithm over to the new lambda syntax (by me). It caused the dmd unit tests to fail though. There is also an issue with intermodule visibility of nested structs that popped up and prevented the every function from being switched over. I should probably recreate
the pull request so one of the compiler guys can take a look.

BA

https://github.com/D-Programming-Language/phobos/pull/744

Reply via email to